Technique for grouping instructions into independent strands

Abstract

A device compiler and linker is configured to group instructions into different strands for execution by different threads based on the dependence of those instructions on other, long-latency instructions. A thread may execute a strand that includes long-latency instructions, and then hardware resources previously allocated for the execution of that thread may be de-allocated from the thread and re-allocated to another thread. The other thread may then execute another strand while the long-latency instructions are in flight. With this approach, the other thread is not required to wait for the long-latency instructions to complete before acquiring hardware resources and initiating execution of the other strand, thereby eliminating at least a portion of the time that the other thread would otherwise spend waiting.

Description

BACKGROUND OF THE INVENTION[0001]1. Field of the Invention[0002]The present invention generally relates to multithreaded programming and, more specifically, to a technique grouping instructions into independent strands.[0003]2. Description of the Related Art[0004]In a multithreaded processing paradigm, a processing unit may execute multiple threads. Those threads may share a hardware resource in order to execute different portions of a multithreaded software program. For example, a first thread could execute using the hardware resource to implement a first portion of the multithreaded program while a second thread waits for access to the hardware resource. When the first thread completes execution, the second thread could then execute a second portion of the multithreaded program using the hardware resource. The hardware resource could be, for example, an execution unit, an arithmetic logic unit, a processing core, or any such hardware resource.[0005]Problems arise with the approach...

AI Technical Summary

Benefits of technology

The patent text claims that the invention improves the performance of a processor by making it more efficient and productive. This means that the processor can work faster and use less energy compared to older processors.

Problems solved by technology

Problems arise with the approach described above when the multithreaded software program involves long-latency instructions, such as load instructions or texture fetch operations.

Consequently, during the time spent waiting for the load instruction to complete, the hardware resource cannot perform any useful work.

In short, the execution of conventional multithreaded software programs fails to efficiently utilize limited hardware resources.

Images